How do I select the right LSE for my STM32L433 when the LSE I have isn't working well?

[LAdam.1]

I used an LSE component proven out on another design containing a STM32F412 for a new design using a STM32L433. The LSE part is Abracon ABS06-32.768kHz-6 which is listed in AN2867 as compatible with the L4 family. I'm using the HAL_RCCEx_EnableLSCO() function to pipe out my LSE signal and it is underwhelming. The duty cycle is roughly 40/60 and while the period is close, it's not as consistence as I would expect. I have tried changing the drive strength with the __HAL_RCC_LSEDRIVE_CONFIG() function. While the raw signal at the crystal looks better, the piped output doesn't improve when adjusting the drive strength. Drive strength is the only difference I can think of between the STM32F412 design and mine as the low drive for it is 1uA and the high drive for the STM32L433 is 630nA. Is there something else I am missing in software or in my crystal selection to explain the difference in performance?

The layout is tight and I also have CSS turned off per the errata notes.

Thanks,

Lauren

[Tesla DeLorean]

What load capacitors are you using?

ST's NUCLEO uses an NDK NK3215SA 32.768 Khz 6pF, with a pair of 6.8pF

That's probably on the high side, depending on what the stray capacitance looks like.

I'd probably try a couple of different values in your board/circuit as built and see if you can improve things that way.

[Bouraoui Chemli]

Hi @LAdam.1​ 

I confirm the ABS06-32.768kHz-6 is compatible with STM32L4.

In deed, you can refer to AN2867, section "3.4 Oscillator transconductance" to calculate gmcrit. Then, check if Gm_crit_max > gmcrit. The Gm_crit_max is provided in STM32L4 datasheet, Table 57 "LSE oscillator characteristics (fLSE = 32.768 kHz)".

Also, you can refer to section "3.3 Load capacitance" to calculate the external capacitors.

What drive mode did you use?

Bouraoui

[LAdam.1]

I have tied every mode level, and they all gave about the same response. technically driving at anything about LOW will meet the gmcrit criteria. When I measure the piped out value of my LSE signal my average frequency is close to expected, but the range of frequencies is very wide at around 32.3kHz-33.3kHz.

[Bouraoui Chemli]

To make sure that Gm_crit_max > gmcrit, I suggest to use Medium low drive mode.

What is the value of the used  external capacitors (CL1 and CL2)?

Bouraoui

[LAdam.1]

My CLs are 6.8pF. I tried lowering them without a noticeable difference in my output signal.

[waclawek.jan]

Tell us more about your hardware. What is the power supply? Are there any power components around? Or any high-frequency circuitry? Can you post the layout of the crystals surroundings?

JW